Python 3.0 was derailed by arrogance that developers should commit to a one-way transition that would touch every function rather than accept that, in Python 3.0, 'x = u"Hello"' should have been a valid statement. It didn't help that it ran slower, added nothing, and broke tools.
Python 3.3 was the first release that had a prayer, but there are mistakes everywhere. For example, virtualenv was included but broke because pip was not. Libraries like ftfy are required because the encodings break. Explaining the oddities of scoping inside generator expressions creates tricky interview questions. And then, Python 3 didn't fix all the broken.
By broken, I mean actually broken. We know where its broken: lots of standard libraries like collections.namedtuple which has an argument to print a generated class. Strange cruft like calendar.weekheader() that only helps one developer's program. This code is in the standard library. Handy things like cleaning up Unicode, DSL support, local encodings, security restrictions. Those you add from other libraries.
Also, where's the love? The courage? I would love to see Python seriously consider dropping case and underscore sensitivity in order to speed up developers, an_item = anItem +1 would be a warning. I would like to see language translation support in the language, great packaging that just works, incorporation unit tests into the package system, reforming the dunder mess, anything! Instead I see the mediocrity by arrogance.
Just for fun, they moved the US Pycon conference to Canada. Only little people have troubles with international travel. Arrogance.